The Korean Telecom Giant at the Center of Anthropic’s Mythos Controversy
Days before Anthropic took its most advanced AI models offline, the White House ordered the company to revoke SK Telecom’s access to Claude Mythos over claims of alleged ties to China.
In a dramatic escalation of national security concerns surrounding cutting-edge artificial intelligence, the Biden administration in June 2026 ordered Anthropic to immediately suspend South Korean telecom giant SK Telecom’s access to its most powerful AI model, Claude Mythos. The directive came just days before Anthropic voluntarily took Mythos—and other frontier models—offline amid a broader reassessment of AI safety protocols.
The controversy centers on SK Telecom, a major investor in Anthropic and the largest telecom provider in South Korea. U.S. intelligence officials flagged SK Telecom’s deep partnerships with Chinese technology firms—including Huawei and state-linked cloud providers—as a potential channel for unauthorized AI model access or data leakage. While SK Telecom has denied any wrongdoing, the White House’s intervention underscores a growing anxiety in Washington about AI proliferation and foreign influence.
The White House Order
According to sources familiar with the matter, the White House National Security Council sent a classified directive to Anthropic in early June 2026, demanding that the company cut off SK Telecom’s API access to Claude Mythos within 72 hours. The order cited “credible intelligence” suggesting that SK Telecom’s Chinese joint ventures could provide Beijing with indirect access to the model’s capabilities.
Anthropic complied, but the decision sent shockwaves through the AI industry. SK Telecom had been one of the first strategic investors in Anthropic, contributing over $100 million in 2024 and securing early access to Claude models for its enterprise cloud customers. The sudden revocation not only strained a key partnership but also raised questions about how far governments might go to control AI distribution.
Anthropic’s Own Pause
Remarkably, Anthropic itself had been planning to take Claude Mythos offline even before the White House’s order. In a blog post published the day after the SK Telecom access was cut, Anthropic CEO Dario Amodei announced that the company would pause development and deployment of all models exceeding a certain capability threshold—including Mythos—pending a comprehensive safety review.
“We have reached a point where the risks of deploying our most advanced systems may outweigh the benefits until we have better safeguards,” Amodei wrote. The decision was reportedly driven by internal research indicating that Mythos could be used to generate sophisticated disinformation or automate cyberattacks with minimal human oversight.
What Is Claude Mythos?
Claude Mythos was Anthropic’s flagship model in 2026, designed to handle complex reasoning, long-context tasks, and multimodal inputs. Unlike earlier Claude versions, Mythos incorporated a novel “constitutional AI” architecture that allowed it to self-correct ethical violations. However, its raw capabilities also made it attractive to state actors seeking to weaponize AI.
Industry analysts note that SK Telecom’s access was particularly concerning because the company had been testing Mythos in telecommunications infrastructure—including 6G network optimization and automated customer service—where any backdoor or data exfiltration could have systemic effects.
Geopolitical Fallout
The controversy has deepened the ongoing rift between the U.S. and South Korea over technology supply chains. South Korean officials have protested the White House’s unilateral action, arguing that SK Telecom is a trusted ally with robust cybersecurity protocols. Some have accused Washington of using national security as a pretext to hobble a foreign competitor.
Meanwhile, Anthropic finds itself caught between its investors and its government. The company has long positioned itself as a responsible AI developer, but the Mythos saga reveals the difficulty of maintaining ethical standards when powerful actors—and powerful governments—are involved.
What Comes Next?
As of late June 2026, Mythos remains offline, and SK Telecom’s access has not been restored. Anthropic has established a joint review board with the U.S. Department of Commerce and the National Security Agency to define new access-control rules for frontier models. The company has also promised to disclose more details about its partnerships, though skeptics argue that true transparency remains elusive.
The episode marks a turning point in AI governance: for the first time, a government has directly intervened to shut down a private company’s AI access to a foreign partner. It may set a precedent for how the U.S. and its allies—or adversaries—handle the next generation of transformative AI systems.
via Wired AI